Image quality standards for cultural heritage digitization projects

Why the quality control is so important?

Digitizing a collection is a Herculean task in terms of labour, costs, and mindshare. It requires a coordinated work of different departments and professionals, and the allocation of a huge amount of institutional resources. Therefore it is essential to be sure about its quality and efficiency.

To achieve good and useful digitization it is not enough to take some nice-looking, high-resolution photos, but the digitization process has to result a standardized, quality-controlled container, including qualified images and metadata: a Preservation Digital Object. A PDO faithfully records the current state of the real-world object, and in the same time, it can be a digital surrogate, and even, for some purposes, the substitution of the original artefacts.[1]

Quality control is not just a simple image check, but it can help the institution to overview and optimize the entire workflow and to verify the adequate and efficient utilisation of the reprographic system. Therefore it might be useful to consult an external expert, and with her/his help, check and improve the process from time to time.


To have a reliable Digital Object (image and data together!), conformity to objective standards is necessary. Most of the collections, archives and museums requir the fulfilment of either FADGI (Federal Agencies Digital Guidelines Initiative) or Metamorfoze Preservation Imaging Guidelines, or the conformity to the international ISO 19264 standard. These institutional standards define image quality with objective, measurable, numerical target values.
In Hungary, the White Paper provides guidance for the digitization of cultural heritage objects.


Both FADGI and Metamorfoze standards describe image quality requirements with several parameters. The most important ones are about the fidelity of color reproduction, the correct rendering of tones, the required resolution and detail transfer. In the same time, the acceptable level of noise and sharpening is maximized.

These sets of requirements define several levels of compliance; the highest is the Preservation level. Both FADGI 4 Stars and Metamorfoze Strict require very high overall image quality to achieve this.

The evaluation process: compare a photo of a suitable test chart (target) with a digitally stored ideal reference, quantifying the differences, then the confirm the compliance.

The Image Science Associates Golden Thread Procedure

There are many kinds of test charts, but for cultural heritage purposes, the most proven one is the Golden Thread system, developed by Image Science Associates. This is a complex tool,  helps us to set up and evaluate the following image parameters: Exposure, Contrast, Resolution, Tone, Color, and Detail rendering, Color Registration Deviation, Flatness (color and tone), Noise, Sharpening, and it has other advanced evaluation features too.

Golden Thread Device Level Target ©ISa

Golden Thread Device Level Target ©ISa

The system consists of three different targets for different purposes and a software for the analysis.

  1. Device Level Target – a test chart to check all parameters of the setup, the camera, the lens, the lighting, etc.
  2. Object Level Target – a test chart which has to be next to each artifact, and
  3. Standard Format Film Targets (monochrome) to set up and verify the digitization of transmissive materials.
  4. An important element of the system is the scientific-grade image quality testing software: the Golden Thread Analysis, but a basic evaluation and certification can also be performed with other programs (Open DICE,

A photograph of the Device Level Target should be taken before and after each production phase, or at the beginning and at the end of each day. That image records all settings, conditions of the reprographic system, hence make the Quality Control possible afterwards.

The Object Level Target must be included on the reproductions next to the object. This tool is the proof of the archiving quality, and it makes possible to numerically check it afterwards on each image.

For example, the L*a*b* values of swatch #15 on this Object Level Target are L62.15, a-1.07, b0.19. So even in 100 years, it will be possible to read it numerically and the viewer will have information about the tones of the original relative to that reading. To make this possible, we need to have the most exact digital representation of the real-world objects.

Golden Thread Object Level Target #15 ©ISa

Golden Thread Object Level Target #15

For us, photographers, it is essential to –at least at a basic level–, understand the science behind the quality control processes. If a correction is needed to fulfil a specific standard, we should know which shooting condition to adjust to get a better result.


The standards also specify the require sharpness and detail rendering expected. Although you can set a high theoretical resolution on the scanners and other capture devices, you also need to measure whether the device is actually capturing the amount of detail that matches the set resolution.

The detail rendering of the reproduction can be quantified by determining the Sampling Efficiency calculated from the MTF (Modulation Transfer Function) or SFR (Spatial Frequency Response) value. This metric objectively determines the differences from an ideally perfect metering chart, taking into account all physical and software elements: the camera, the sensor, the lens, the lighting, and the export settings of the RAW workflow.

The FADGI 4 Stars (preservation-grade) requirement is also strict in this: MTF10 (10% SFR) requires> 90% sampling efficiency.

>>> The method of measuring the resolution and evaluation of detail rendering, will be discussed in the next part of the article >>>

DeltaE explained

The software determines the color rendering by measuring the difference between the color swatch on the reproduced target and the digitally stored reference color. The Delta E 2000 method, developed by the CIE (Commission Internationale d’Éclairage), an international organization for the scientific description of colors, uses a complex algorithm to calculate and quantify the color difference.

The FADGI 4* specification allows a very small color difference (ΔE2000 <3), thus guaranteeing objective color accuracy.

The online service analyzes and evaluates the following main properties: Color Deviation (ΔE00), Sampling Efficiency (SEFF), Color Channel Slip (CMR), MTF 10 and 50, MTF max, Effective Resolution (DPI), Brightness Deviation (μL*), Illumination Unevenness (ΔL*), SFR High and Medium Frequencies, Oversharpening, Color Coding Error (ΔE 2000), Total Noise and Color Channel Misregistration (RGB deviation)

In addition, this online service qualifies the findings according to the FADGI standard (4 stars is given for maximum compliance).

The shooting conditions only need to be set once, and as long as one of the parameters does not change, images that comply with the digitization standard are guaranteed to be taken. evaluation panel


Image quality control is an essential element of cultural heritage digitization. The ISa Golden Thread procedure examines every element of the entire workflow, so quality control with this tool truly guarantees the adequacy of digitization.
Even though its theoretical background is very complex and it is based on the results of decades of scientific research, the practice of the quality control process is rather simple when using the right tools, and it also can be mastered with a short training.

Resources, links, further readings


If you have any further questions, please contact me through the contact form: CONTACT 

Daniel Horvath
freelance photography technology consultant


[1] „The Ultimate Goal: Digital Surrogacy. A PDO is not just a digital copy of the original object. It is a complete package ready to function as a digital surrogate for the original physical object in as many use-cases as possible. A PDO is well structured, easy to access and understand, is well described, and contains high quality content. This means that a requesting entity (patron, researcher, administrator, marketing agent) who wishes to have access to the original object can be instead be offered access to the PDO, with no noticeable difference in value to that entity.” Source: